Upload
trinhthuan
View
222
Download
0
Embed Size (px)
Citation preview
13th World Conference on Earthquake Engineering Vancouver, B.C., Canada
August 1-6, 2004 Paper No. 1089
LIQUEFACTION STRENGTH OF COARSE-GRAINED SOILS AS DETERMINED BY LARGE TORSIONAL TEST AND
THE MEMBRANE PENETRATION EFFECT ON THE STRENGTH
Yasuo TANAKA Shigetoshi SUGIMOTO Naoyuki ADACHI 1
SUMMARY The liquefaction strength of coarse-grained soil was investigated by using a large torsional hollow cylinder test and a medium size triaxial test, and the membrane penetration effect on the strength during the liquefaction test was also investigated. The results of liquefaction tests were discussed with the following two aspects; 1) The applicability of existing strength correction method for membrane penetration effect that uses the test results without the membrane penetration correction. 2) The difference between the liquefaction strengths that are obtained from the hollow cylinder test and the triaxial test. As to the applicability of existing strength correction method for the membrane penetration effect, it was concluded that the existing method underestimates somewhat the strength in both type of tests. For the comparison of the liquefaction strengths between the two types of test method, it was found that the liquefaction strength obtained from the large torsional hollow cylinder test is higher than that from the medium triaxial test. Review is made on the past studies of liquefaction strength difference between the hollow cylinder and triaxial tests on clean sands, and it was shown that the existing method to convert the triaxial test result to that of torsional hollow cylinder test based on the studies with clean sand is not applicable to the coarse-grained soil tested in this work.
INTRODUCTION Liquefaction strength of soil has been studied extensively in the past, but the experiments have been mainly for clean sands as the many of liquefied sites are located at coastal areas or alluvial deposits consisting of clean sand. But in Hansin-Awaji Earthquake, not only the clean sand but also the coarse-grained soil, especially of fill material at man-made islands near Kobe-port1), liquefied extensively causing a heavy damage to harbor facilities of Kobe Port. Because of such liquefaction of fill material that contains a large size aggregates and with a well grade characteristics, investigations on the liquefaction characteristics of coarse-grained soil became one of the major research item of soil liquefaction. The liquefaction characteristics of coarse-graind soil have been investigated in our laboratory,
1 Kobe University Research Center for Urban Safety and Security
and a special apparatus of large torsional hollow cylinder test2) was constructed to simulate
the deformation condition that prevails during the earthquake in the field. Usually the
liquefaction strength of coarse-grained soil is studied by performing a large size triaxial test,
but it is known that there is a difference between the strengths obtained from the triaxial
test and the torsional hollow cylinder test. The strength difference between the two types of
test may be due to these two different modes of shearing with respect the soil structure or
the orientation of soil depositional direction, and the anisotropic properties of soil would play
a major role in this aspect of soil behavior. The difference of soil strength among the
different types of soil testing has been studied mostly on static strength properties of soil,
but there is a need to correlate the liquefaction strengths obtained from these two types of
test as the use of triaxial test would be more conventional for usual soil investigations.
Another problem associated with the liquefaction strength of coarse-grained soil is the need
to consider the membrane penetration effect. Membrane penetration occurs in a great extent
for the specimens consisting of large size particles. It is well known that the membrane
penetration effect increases with the mean diameter of soil particle3). Several studies have
been made to correct the membrane penetration effect on the liquefaction strength, either
directly correcting during the test or correcting indirectly by applying a correction factor on
the test result that was obtained without membrane correction. In this study, a special
device was used to correct membrane penetration effect during liquefaction test, and the
difference of corrections between the direct and the indirect methods of correcting the
membrane penetration effect on the liquefaction strength was examined.
EXPERIMENTAL PROCEDURE
Test Apparatus Fig. 1 shows a large hollow cylinder test apparatus developed herein. As to the size of
specimen, outside and inside diameters of the specimen were 50cm and 30cm, and the height
was 60cm. The axial force to specimen is given through the piston via the air cylinder placed
above the apparatus, and the torsional force to the specimen is given by two horizontal arms
connected to the air cylinders placed at upper edge of outer circular frame. The air pressure
to these cylinders was controlled by electro-pneumatic regulators which were in turn
operated by personal computer. Measurements of axial and torsional forces on the specimen
were made by a two-directional-load-cell that is placed just above the specimen cap, and cell
pressure and pore pressure were measured by pressure transducers. Measurements of axial
displacement and volume change of the specimen were made by LVDT and load cell to
measure the volume of water expelled from the specimen. A membrane penetration
correction device is attached at the drainage line from the specimen, and its function will be
described in details later.
The medium size triaxial test apparatus is seen in Fig. 2. The diameter of the specimen was
10cm and its height was 20cm. Loading and measurement systems are almost the same with
those of the large torsional hollow cylinder test apparatus. In case of the triaxial test, the
volume change was measured by an electronic balance for better accuracy and the
membrane correction devise with finer adjustment was used as will be shown later.
0
10
20
30
40
50
60
70
80
90
100
0.001 0.01 0.1 1 10 100Soil particle diameter(mm)
Nishi-soil
Silica sand
Fig.3 Gradation curve
①
② ③
④
④⑤
⑩
⑫
1
3 45
6
2
P
01
3 45
6
2
01
3 45
6
2
P
P
01
3 45
6
2
P
⑥⑦
⑧
⑨
⑪
amp
A/
Dcard
①air cylinder(for axial stress)
②rotation gauge
③LVDT
④air cylinder@2(for shear stress)
⑤arm for shear stress
⑥two-direction load cell
⑦cell
⑧hollow cylinder specimen
⑨inner hole
⑩flame(for shear stress)
⑪pressure gauge(for pore pressure)
⑫pressure gauge(for dell pressure)
⑬volume change measure device
⑭electric regulater(for axial stress)
⑮electric regulater(for sher stress)
⑯membrane penetration correcting device
⑬Dout:500m
Din:300mm
H:600mm
01
3 45
6
2
01
3 45
6
2
P
P
⑭
⑯
⑮
analog voltage
degital
voltage
personal
computer
degital
voltage
amplify
Fig.1 Large hollow cylinder test apparatus Fig.2 Medium size triaxial test apparatus
Table 1: Physical Properties of material
material Nishi-soil Silica sand
ρs(t/m3) 2.67 2.62
ρmax(t/m3) 1.95 1.57
ρmin(t/m3) 1.54 1.25
Dmax (mm) 19 0.85
Dmean (mm) 1.7 0.22
Uc 15 2.0
Uc' 0.82 1.1
Test Materials and Sample Preparation The experimental material was a coarse-grained gravelly sand taken from a construction site at Nishi-Ward of Kobe City (hereafter it will be called as Nishi-soil). The maximum grain size of the sample was adjusted to 19mm. The gradation curve is shown in Fig. 3, and it is seen that the Nishi-soil is well-graded material. The physical properties of the soil are shown in Table 1. Sample preparations for both the large
hollow cylinder and the medium size
triaxial tests are the same as described in
the followings. All specimens were
prepared by pulverizing soil into a mold
and the density of soil was set to a relative
density of 60% by adjusting falling height
of air-pulverization. For saturating the
specimen, a specimen was given an ample
amount of CO2 flow from the bottom, while
maintaining a confining pressure of
0.04Mpa to the specimen. Then enough
flow of de-aired water was introduced into
the specimen, and then a back-pressure of
0.1Mpa was given to the specimen to
increase the degree of saturation. Before
the test, the measurement of “B” value was
made for each specimen, and only those specimens with B value of greater than 0.95 were used
for the liquefaction test. It may be noted that all specimens were consolidated to the effective
confining pressure of 0.1MPa that was achieved by applying a cell pressure of 0.2MPa with a
back pressure of 0.1MPa.
Liquefaction Strength Test In liquefaction test, a sinusoidal cyclic of shear stress was applied to the specimen in undrained condition. In the case of large torsional hollow cylinder test, a sinusoidal cyclic shear stress was applied by giving a rotational twist to the specimen while keeping the isotropic confining stress. A maximum shear stress is applied on horizontal plane of the specimen. In the case of the triaxial test, a sinusoidal cyclic deviator stress was given by increasing and decreasing the axial stress and the maximum shear stress acts on 45°plane from the horizontal. The large torsional hollow cylinder test was terminated when the excessive pore pressure exceeds 95% of the confining pressure, while the triaxial test was terminated when the excessive pore pressure exceeds 95% limit and DA exceeds 5%.
PROCEDURES FOR CORRECTING MEMBRANE PENETRATION
Correction of membrane penetration The amount of membrane penetration (MP) to be corrected for hollow cylinder sample and triaxial sample is obtained by the following procedure suggested by Vaid & Negussey4). As indicated below, the volumetric strain of soil skeleton is assumed to be three times of soil axial strain that is measured in the unloading procedure:
aV εε ×= 3
100V
s VVε
×=∆
aε
:axial strain(%)
Vε
:volumetric strain of soil skeleton(%)
sV∆ :volume change of soil skeleton (cm3)
V :initial volume(cm3) Measured volume change is composed of not only the volume change of soil skeleton but also of the change due to MP. Therefore change of membrane penetration can be obtained by subtracting the volume change of soil skeleton from the measured volume change:
sVVMP ∆−∆=
MP:amount of MP (cm3) V∆ :measured volume change (cm3)
By applying the above method, the relationship between MP and the effective confining pressure, p’, can be obtained. Since the amount of MP correction should be unique for the soil with same gradation, a single correction curve for MP and p’ is determined by averaging the measured MP vs. p’ relationships. Fig.4 shows the obtained correction curves for both
b)
a)
handle
piston
cylinder
L
V
D
T
pore water
vertical
displacement
rotational displacement
Fig.5 MP device (Hollow cylinder test)
types of tests. Other system compliances In addition to the MP correction, there are two other system compliances to be corrected for the test apparatus. One is the system compliance of drainage tube and the other is that of MP correction cylinder, both being function of confining pressure changes. It was found that these system compliances are much smaller than that of MP of specimen
5),
6), but these were also corrected by using the MP correction
device during liquefaction test.
METHOD TO CORRECT MP Direct correction method The MP device used in the large torsional hollow cylinder test is shown in Fig.5 and one used in the triaxial test is shown in Fig.6. In both devise, a minute amount of water is injected in or out of the specimen by horizontal displacement of piston which is generated by rotation of mega-torque motor through a screw gear5), 6). The amount of injected water is the product of cylinder’s section area and the piston displacement which is measured by LVDT. By applying the measured p’ value on the MP correction curve of Fig.4, the amount of water to be injected is calculated and then the MP correction device is operated to give the necessary volume of water. Fig.7 shows a schematic of automatic MP correction procedure in the triaxial test. In the triaxial test, the accuracy of correction was 0.0088% of the maximum amount of MP. On the other hand, the MP correction device for hollow cylinder test is operated manually5).
Fig.4 Correction Curves for Membrane penetration effect, a) large torsional hollow cylinder test, b) medium size triaxial test
Fig.6 MP device (Triaxial test)
Pressure
gauge
Ideal M P
quantity
R eal M P
quantity
M P device
M egatorque
m otor
C om pare
R eal>Ideal
Extraction
R eal<Ideal
Injection
C orrecting
Test
epuipm ent
Fig.7 Correction procedure( Triaxial test)
Fig.8 CN vs.CRM curve7)
Indirect correction method As already stated, there is an indirect MP correction method which modifies the liquefaction test result without MP correction. Martin et al.7) has proposed a correction coefficient of CRM which is a relationship between MP and the volume change of soil skeleton. Nakamura et al.8) has studied the liquefaction strength of various types of soils whose by using a MP correction device developed by Tokimatsu et al.9). From test results, they proposed a strength correction curve by using CRM value as shown in Fig.8. In the figure, the vertical axis, CN=R/R*, represents a correction ratio to be applied on the liquefaction strength curve that is obtained without MP correction. For the same level of cyclic stress, the liquefaction strength curve without correction yields the number of cycles R, while the strength curve with MP correction yields the R* cycles. Since the liquefaction strength is overestimated with the tests without MP correction, the liquefaction strength curve should be shifted horizontally towards the origin by the ratio given as CN.
Table 2 Hollow Cylinder Test Series Table 3 Triaxial Test Series
H series τc/σ
c' τ
cD
r(%)
H-0.35 0.35 0.035 62.82
H-0.25 0.25 0.025 54.01
H-0.20 0.20 0.020 64.46
H-0.15 0.15 0.015 62.18
H-0.35-MP 0.35 0.035 62.14
H-0.25-MP 0.25 0.025 63.44
H-0.20-MP 0.20 0.020 62.76
H-0.15-MP 0.15 0.015 59.29
T series σd/2σc' σd Dr(%)
T-0.30 0.30 0.060 59.52
T-0.25 0.25 0.050 57.56
T-0.20 0.20 0.040 62.08
T-0.15 0.15 0.030 61.86
T-0.30-MP 0.30 0.060 59.32
T-0.25-MP 0.25 0.050 62.02
T-0.20-MP 0.20 0.040 61.55
T-0.15-MP 0.15 0.030 59.24
TEST RESULTS Tables 2 and 3 show the torsional hollow cylinder and the triaxial test series performed in this study. As can be seen from the tables, four tests in each test series were performed by either applying the MP corrections or without the correction.
Undrained Cyclic Loading Test Results With or Without MP Correction Hollow Cylinder Test Series Fig.10 shows typical results of liquefaction test by using hollow cylinder test apparatus, Test [H-0.20] for one without MP correction and Test [H-0.20-MP] for one with MP correction. In correcting the MP effect, the test [H-0.20-MP] is performed by manually adjusting the correction device, and the obtained effective stress path shows some scattering. It can also be seen that the number of cycles to liquefy the specimen in Test [H-0.20-MP] is much less than that of Test [H-0.20], and thus the effect of MP on liquefaction strength is clear demonstrated. Triaxial Test Series Fig.11 shows typical results of liquefaction test by using medium size triaxial test apparatus, Test [T-0.20] for one without MP correction and Test [T-0.20-MP] for one with the MP correction. In contrast to the manually controlled adjustments for the membrane correction for the hollow cylinder tests as above, automatic adjustment of membrane correction device is used for the triaxial tests, and the stress path of Test [T-0.20-MP] is very smooth compared with the one for hollow cylinder test, for example Test [H-0.20-MP]. Thus it can be concluded that the automatic correction of MP is very effectively achieved. It can also be seen that the number of cycles to liquefy the specimen in Test [B- 0.20-MP] is much less than that of Test [B-0.20], and therefore the effect of MP on the liquefaction strength is also significant in the medium size triaxial test series.
Fig.10 Large Hollow Cylinder Test Results, a) Test [H-0.20], b) Test [H-0.20-MP]
a) shear stress vs.time(Test: H-0.20) a) shear stress vs.time(Test: H-0.20-MP)
b) pore water ratio vs.time (Test: H-0.20-MP)
c)γ vs.time(Test: H-0.20-MP)
d)τ vs.γ (Test: H-0.20-MP)
e) stress path (Test: H-0.20-MP)
b)
b) pore water ratio vs.time(Test: H-0.20)
c)γ vs.time(Test: H-0.20)
d)τ vs.γ (Test: H-0.20)
e) stress path (Test: H-0.20)
a)
Fig.11 Medium Size Triaxial Test Results, a) Test [T-0.20], b) Test [T-0.20-MP]
b) pore water ratio vs.time(Test: T-0.20)
a) shear stress vs.time(Test: T-0.20)
c)γ vs.time(Test: T-0.20)
d)τ vs.γ (Test: T-0.20)
e) stress path (Test: T-0.20)
a)
a) shear stress vs.time(Test: T-0.20-MP)
b) pore water ratio vs.time (Test: T-0.20-MP)
c)γ vs.time(Test: T-0.20-MP)
d)τ vs.γ (Test: T-0.20-MP)
e) stress path (Test: T-0.20-MP)
b)
Fig.12 Liquefaction strength curve (H series)
Fig.13 Liquefaction strength curve (T series)
Fig.14 Ratios plotted on CN - CRM curve7)
Liquefaction Strengths Curves With or Without MP Correction Fig.12 shows a comparison of the liquefaction strength curves of those tests with or without MP corrections based on the hollow cylinder tests. Similarly Fig.13 compares the liquefaction strength curves of those test results for the triaxial tests. As can be seen from these figures, there is a significant difference between the strength curves of with and without MP correction, and the horizontal distance between these curves are used to express the amount of correction needed to be applied to the test result obtained without MP correction. Also shown in Fig.12 and Fig.13 are the liquefaction strength curves that were obtained from the indirect correction of the test results without MP correction. These curves were obtained by applying CN - CRM curve as already shown in Fig.9. Calculated CRM was 0.64, and corresponding value of CN of 2.6 was obtained for correcting the number of cycles for the torsional hollow cylinder test. For the triaxial tests, CRM was 1.43 and corresponding value of CN was 3.0. From these figures, it is shown that directly corrected liquefaction strength is slightly higher than the ones with indirect correction irrespective of different testing methods. Therefore the correction curve as obtained from Fig.9 is not applicable to the coarse grained soil tested herein. Based on the strength curves as obtained herein using the MP correction device, the values of CN was reevaluated, and it is 1.43 for hollow cylinder test and 2.27 for triaxial test. These two CN values have been plotted in Fig.14. As shown in the Figure, both values locate below the line determined by Martin et al.7). It also
appears that CN value of hollow cylinder test locates slightly lower than that of triaxial test.
Fig.15 Two liquefaction strength curves
10)
Fig.16 Definition ofεa comp/DA
10)
Fig.17 {(τc/σc’)/(σd/2σc’)}
- (εa comp/DA) curve8)
Fig.18 Liquefaction strength curve
(H-MP&T-MP series)
Comparison of liquefaction strength between the large torsional hollow cylinder test and the medium triaxial test As have been shown previously, there is a significant difference in the direction of shearing relative to the bedding plane of specimen. The anisotropy of soil fabric would influence the deformation and the strength properties of soils. Toki et al.10) have studied the difference of liquefaction strength between hollow cylinder test by noting the effect of soil fabric orientation on the liquefaction strength and they have examined the variations in liquefaction strength among the clean sand specimens that have been prepared by various specimen preparation methods. An example of Toki et al. results is shown in Fig.15. They concluded, based on the various sample preparation methods and strength difference between the two types of test, that the ratio of normalized strength {( τ c/ σ c’)/( σ d/2 σ c’)} between hollow cylinder and the triaxial tests depends on the degree of fabric anisotropy. They have expressed the degree of the anisotropy by a strain ratio between compression strain (εa comp) and double amplitude strain (DA) both measured during the triaxial testing. Fig.16 illustrates the definition of these strains, and the soil fabric response is isotropic when the ratio is 1/3. Thus they have demonstrated the strength difference between the two tests ( i.e., {(τc/σc’)/(σd/2σc’)}) varies with the anistropic fabric parameter ( ε a comp/DA), and their results are shown in Fig.17 by plotting the results on {(τc/σc’)/(σd/2σc’)} - (εa comp/DA) coordinates. Toki et al. results (Fig.17) which is basically for clean sands is used to compare the
results from the two types of tests in this study. The value of (εa comp/DA) from [T-MP] series was about 0.47 for the coarse grained soil in this study and thus the material shows nearly an isotropic deformation response. The correspond value of {(τc/σc’)/(σd/2σc’)} as suggested by Toki et al. is about 0.8, and this estimated strength difference is depicted in Fig.18. As can be seen in Fig.18, the liquefaction strength difference as obtained from this study is much less than that of Toki et al. study. It is quite possible that the difference of soil type (gradation or grain size) would influence the anisotropic response of soil. Further investigation is needed on how the difference of soil type would affect the anisotropic response such as (εa comp/DA) value. However, it is clear that the coarse-grained soil such as Nishi-soil behaves differently from the behavior of clean sand.
CONCLUSION Based on the above test results, the following conclusions may be obtained for the liquefaction strength of coarse-grained soil like Nishi-soil that: 1) The effect of MP on the liquefaction strength of coarse-grained soil is very large
irrespective of the different types of testing. 2) Existing indirect correction method of MP effect on the liquefaction strength somewhat
overestimates the reduction of the liquefaction strength that is obtained from the tests without MP correction.
3) The difference of the liquefaction strengths of coarse grained soil as obtained from the torsional hollow cylinder and the triaxial tests is much smaller than the difference known for clean sand. Further investigation is needed on how the difference of soil type would affect the difference of liquefaction strengths as obtained from these two types of test.
REFERENCES 1. Yasuo TANAKA, Kouichiro YAMADA, Noritaka HASEGAWA, Tomoaki HIRAOKA.
“ Liquefaction characteristic of fill material used in Kobe-port-island and Rokko-island. “ Construction engineering laboratory report, 1996, 38-B: 97-106.
2. Shigetoshi SUGIMOTO, “ Decision of liquefaction strength used the large torsional
hollow cylinder test and effect of membrane penetration.” A master’s thesis of graduate school of Kobe Univ, 2003.
3. Frydman S, Zeitlen J G, Alpen I, “ The Membrane Effect in Triaxial Testing of Granular
Soils” , Journal of Testing and Evaluation, 1973, Jan, Vol.1, No.1: 37-41. 4. Yoginderl P Vaid, Dawit Negussey, “ A Critical Assessment of Membrane Penetration in
the Triaxial Test”, American Society for Testing and materials, 1984: 70-76,. 5. Naoyuki ADACHI, “ Membrane penetration correcting used liquefaction test used the
large torsional hollow cylinder test “ A graduation thesis of graduate school of Kobe Univ, 2003.
6. Koji TONEGAWA, “ Relationship between liquefaction strength and membrane
penetration for coarse-grained soil used the medium triaxial test equipment “ , A graduation thesis of graduate school of Kobe Univ, 2004.
7. Martin G R, Finn, W D L, Seed H B, “ Effects of system compliance on liquefaction tests
“ , Journal of Geotechnical Engineering Division,ASCE, 1978, Vol.104,No.GT4: 463-479.
8. Koji NAKAMURA, Koji TOKIMATSU, Yoshiaki YOSHIMI, “ Convenient correcting
method of membrane penetration for undrained test “ , 23’th conference of Geotechinical engineering foundation,1988: 657-658.
9. Kohji TOKIMATSU, Kohji NAKAMURA, “ A liquefaction test without membrane
penetration effects “ , Soil and Foundations, 1986, Vol. 26, No.4: 127-138. 10. Satoshi Yamashita Yousuke TOKI “ Comparison of liquefaction strength based on cyclic
triaxial test and torsional hollow cylinder test “ , 26’th conference of Geotechinical engineering foundation,1991: 725-728.